Hi there
can any one explain to me if we can combine two dimensional reduction methods to reduce more features? is there specific conditions to do that? (for example NPE and PCA)
Yes of course, you can combine two dimensionality reduction methods of the same type, i.e., transformation methods or feature selection methods. In fact, you can fuse the obtained features with NPE and PCA using any fusion method such as, dempster shafer's theory. Good Luck
Yes, many methods were examined such as DWT&PCA, DCT&PCA, PCA&LDA, PCA&SVM,.....etc.
Yes. you can do it. But, this depends on your intended method and application.
For example, sometimes to reduce the size of a problem, it is better to reduce the dimensions of the data with different per-process methods in different steps of the main proposed method, so that it ultimately reduces the number of features and the computational time.
The following article is suggested for a better review:
A new approach for rotation-invariant and noise-resistant texture analysis and classification, M. M. Feraidooni and D.Gharavian, Machine Vision and Applications, Dec 2017.
However, in general, you can use the following combinations:
Different DWT's, DWT&PCA, DCT&PCA, DWT&SVM and ... .
thank you Mohammad Mahdi Feraidooni and Mohammadreza Mohammadrezaei
- Badges
- Science topic